I have extensive experience in robust statistical methods to study complex heterogeneities (referring to differential effects or association) arising from longitudinal data. My recent work suggested the potential of such methods for advancing the research on macrovascular complications/cardiovascular diseases (CVD) in type 2 diabetes (T2DM). My career goal is to develop an independent research career in statistical methods pertinent to the design and testing of interventions aimed at reducing diabetes-related CVD for patients with T2DM. The primary purpose of this proposal is to support an integrated program of training and research to achieve this goal. My career development objectives are to: 1) acquire an in-depth understanding of T2DM regarding its diagnosis, etiology, pathophysiology, and treatment methods and the associated pharmacological theories;2) acquire an in-depth understanding of T2DM related CVD, especially regarding their etiology, diagnosis, and prevention theories;3) gain insights from the health services research perspective about factors that may influence providers'implementation of evidence-based guidelines and patients'choices of treatment and adherence behaviors;4) enhance my statistical skills for developing efficient and robust estimation methods;and 5) enhance my skills for communicating quantitative methods and analytic results with clinicians, clinical researchers, funding agencies, and health policy analysts. The objectives will be met through formal coursework, dyadic mentoring, consultation, seminars and conferences on diabetes and CVD research as well as in Bayesian statistics methodology. For the research proposed, I will apply the general latent variable modeling technique to examine the differential relationship between glycemic control and CVD due to medication adjustments for hyperglycemia, dyslipidemia, and hypertension in a longitudinal clinical cohort of type 2 diabetics. For the purpose of identifying effective strategies for prevention of CVD, I will further develop my statistical expertise to obtain efficient estimates of the differential medication effects on CVD risk, which are mediated through improving glycemic control, hyperinsulinemia, or insulin sensitivity. This research will take advantage of my access to a rich set of longitudinal data available in my current research environment in the VA and the experience of my mentors. Findings of the study will provide further insights about why the field has not yet reached a firm conclusion about the association between "tight glucose control" and CVD, the leading cause of morbidity and mortality in type 2 diabetics. Specifically, the results will inform how CVD intervention strategies should vary across individuals with different characteristics.